Source code for tensorplay.distributions.logistic_normal

# mypy: allow-untyped-defs

from tensorplay import Tensor
from tensorplay.distributions import constraints, Independent
from tensorplay.distributions.normal import Normal
from tensorplay.distributions.transformed_distribution import TransformedDistribution
from tensorplay.distributions.transforms import StickBreakingTransform


__all__ = ["LogisticNormal"]



[docs]
class LogisticNormal(TransformedDistribution):
    r"""
    Creates a logistic-normal distribution parameterized by :attr:`loc` and :attr:`scale`
    that define the base `Normal` distribution transformed with the
    `StickBreakingTransform` such that::

        X ~ LogisticNormal(loc, scale)
        Y = log(X / (1 - X.cumsum(-1)))[..., :-1] ~ Normal(loc, scale)

    Args:
        loc (float or Tensor): mean of the base distribution
        scale (float or Tensor): standard deviation of the base distribution

    Example::

        >>> # logistic-normal distributed with mean=(0, 0, 0) and stddev=(1, 1, 1)
        >>> # of the base Normal distribution
        >>> # xdoctest: +IGNORE_WANT("non-deterministic")
        >>> m = LogisticNormal(tensorplay.tensor([0.0] * 3), tensorplay.tensor([1.0] * 3))
        >>> m.sample()
        tensor([ 0.7653,  0.0341,  0.0579,  0.1427])

    """

    arg_constraints = {"loc": constraints.real, "scale": constraints.positive}
    # pyrefly: ignore [bad-override]
    support = constraints.simplex
    has_rsample = True
    # pyrefly: ignore [bad-override]
    base_dist: Independent[Normal]

    def __init__(
        self,
        loc: Tensor | float,
        scale: Tensor | float,
        validate_args: bool | None = None,
    ) -> None:
        base_dist = Normal(loc, scale, validate_args=validate_args)
        if not base_dist.batch_shape:
            base_dist = base_dist.expand([1])
        super().__init__(
            base_dist, StickBreakingTransform(), validate_args=validate_args
        )

    def expand(self, batch_shape, _instance=None):
        new = self._get_checked_instance(LogisticNormal, _instance)
        return super().expand(batch_shape, _instance=new)

    @property
    def loc(self) -> Tensor:
        return self.base_dist.base_dist.loc

    @property
    def scale(self) -> Tensor:
        return self.base_dist.base_dist.scale